DNA Barcoding of Lepidoptera Species from the Maltese Islands: New and Additional Records, with an Insight into Endemic Diversity
Round 1
Reviewer 1 Report
Review: DNA barcoding of Lepidoptera species from the Maltese Islands: new and additional records, with an insight on endemic diversity.
The work received reports the 1st DNA barcode analysis of Lepidoptera species sampled from Maltese islands. Results indicate a high proportion of novel DNA barcodes not reported for conspecifics sampled elsewhere, indicating the relative uniqueness of intraspecific diversity at the islands. The work uses a standard but uncritical and ofttimes outdated approach to DNA barcode analyses, there are issues with the analyses, reporting and interpretation of results. Much of the caveat literature over the last decade concerning erroneous use of DNA barcode statistics has been overlooked and needs to be considered here.
Table 1: the authors report "Mean % ID" as the "mean percentage identity with species" this is inappropriate. Caveats in DNA barcode papers frequently indicate report of "mean intraspecific" values have limited worth and are especially misleading when considering the more informative statistics of the maximum genetic distance in a BIN or species VS the minimal genetic distance to the nearest neighbour species (as a direct means to establish if a DNA barcode gap exists between genetic sister species). It would be more informative to report the "Maximum percentage difference (Max Intraspecific D)" as an indication of the upper limits of genetic variation exists within the species. There is a diverse library of papers discussing the "DNA barcode gap" and correct the statistics used to explore this issue, and some reporting of this in the Results would be normally expected of a DNA barcode paper
BINs are 1st mentioned in the introduction and later reported in the Results & Discussion, but a simple description of what these are, and why they are useful in analyses, and why they are used in your study, is not presented in the draft. Logically this description should be at least reported in the Methods (and potentially with minor passing comments in the Introduction where 1st mentioned).
NJ genetic distance tree and phylogenetic inference reported in the paper: Authors extensively report NJ distance trees in their results, this is standard and widely used in DNA barcode analyses due to the simplicity of the approach and its relative utility for exemplifying terminal species units based on the genetic distance data. But the authors go one step further and report these NJ relationships as phylogenetic relationships. NJ trees are phenograms (dendrograms) of pairwise genetic distances estimated among terminal samples. Over several decades authorities have argued that NJ distance trees are not appropriate for inference of phylogenetic relationships among terminal samples. The primary reason: NJ combines distances obtained from informative characters (in this case nucleotide sites) into a single ensemble metric - NJ cannot account for individual character (in this case nucleotide) homologies that contribute to the reconstruction and inference of a systemic relationship. There is much discussion of this in the literature, and authors should explore this issue, start with DeSalle R and Goldstein P (2019) Review and Interpretation of Trends in DNA Barcoding. Front. Ecol. Evol. 7:302. doi: 10.3389/fevo.2019.00302. Alternatively, there are a minority of researchers who maintain NJ trees are interpretable as phylogenetic reconstructions. If you can cite or provide a justification for your NJ distance trees to be considered as phylogenetic reconstructions, then report that in your Methods. Otherwise, you will need to change all your reported terminologies involving your NJ trees. For example, "phylogenetic relationship" will need to be replaced by "genetic distance relationship"; mention of "clade" will need to be replaced by "cluster" or "node" depending on context used; "phylogram" replaced by "dendrogram"; "phylogenetic tree" replaced by "genetic distance tree" or “phenogram” etc.
Terminology: In Results section, Monophyly is reported several times in relation to the NJ trees. Despite the above caveats concerning use of NJ trees for phylogenetic inference, it is still acceptable when reporting NJ results to use terms such as "monophyly" when referring to a cluster of specimens which share the same taxonomic description. But I am curious why other similar terms regarding species systematics relationships have not been used in the draft. In almost every subset tree examined there is at least one species which is paraphyletic with respect to a sister species, these are not formally reported as "paraphyletic", rather we are provided with more verbose description that are difficult to follow. For example, in reporting of Fig 3, Nyctobrya muralis has multiple genetic clusters that are poorly resolved in the NJ tree, the species is genetically paraphyletic with respect to N. segunai.
Use of 2% sequence criteria for species delimitation: use of a universal 2% distance threshold as criteria for demarcating species, is cited several times in the paper (eg., lines 191-199) and justification for this is based on reports in several papers cited by authors at line 139. But note this issue of universal distance thresholds applied to DNA barcode studies has been EXTENSIVELY discounted by leading authorities in the field including DNA barcoding founders Hebert at al, and the general consensus is that use of a universal set % threshold for species delimitation is prone to false negative and false positive errors due to differing mtDNA coalescent rates and depths among species – new species do not universally evolve at a set uniform rate of genetic divergence as is naively assumed by use of a sequence distance threshold. Essentially a universal threshold is redundant as an empirically applied analytical statistic in DNA barcode studies (refer summary of issue in Collins RA, Cruickshank RH. The seven deadly sins of DNA barcoding. Mol Ecol Resour. 2013 Nov; 13(6):969-75. doi: 10.1111/1755-0998.12046. Epub 2012 Dec 27. PMID: 23280099). So, what is required here is a more cautious use of language nomination of "potentially" novel taxa, where you have large genetic distances between taxa, and additional supportive evidence of morphological character differences, that is the realm of integrative taxonomy used for species delimitation and discovery and there are plenty of useful paper which describe this approach. Authors should not be so forthright with asserting language such as “justifies it being promoted to species level” used at line 198.
K2P (Kimura 2 Parameter) is used to modify pairwise genetic distances in NJ analyses... why did you do this in a DNA barcode study??? (Refer again to the Collins & Cruishank paper and others)
In your NJ reporting of pairwise comparisons, be careful using "divergence" when you are actually referring to "distance", in systematics work the former suggests reproductive isolation the latter is non-inferential in regard to reproductive isolation.
Minor Comments
lines 149-150: mention of a BOLD generated distance tree used for inferring Results, but not observable in the draft. Either include it in a Supplementary section, or refer to it as "(data not shown, eg., at line 285 and elsewhere).
lines 169-170: report of nucleotide content. I do not see a need for this, it is not reported elsewhere in the draft and has no explained bearing on the outcomes of your work. Normally we would see such reports in comparisons of different loci - or if there is an extreme nucleotide bias not observed relative to earlier studies or of other taxonomic groups. It has no justified reason to be in your draft.
Lines 180-182: There is comment stating 46% of Maltese island barcodes are novel in regard to mainland conspecifics (Table 1). But I cannot determine in your Table 1 which GenBank accessioned sequences are novel. So, I cannot determine how you came up with a 46% novelty statistic. This could be resolved by simply adding an asterisk * to those accessions which are novel, simple and accountable.
Line 322: replace “morphological identification data” with “taxonomic descriptions”
Author Response
Reviewer 1:
We would like to thank reviewer 1 for highlighting a number of points that result in a better interpretation of the data.
Comment: Table 1: the authors report "Mean % ID" as the "mean percentage identity with species" this is inappropriate. Caveats in DNA barcode papers frequently indicate report of "mean intraspecific" values have limited worth and are especially misleading when considering the more informative statistics of the maximum genetic distance in a BIN or species VS the minimal genetic distance to the nearest neighbour species (as a direct means to establish if a DNA barcode gap exists between genetic sister species). It would be more informative to report the "Maximum percentage difference (Max Intraspecific D)" as an indication of the upper limits of genetic variation exists within the species. There is a diverse library of papers discussing the "DNA barcode gap" and correct the statistics used to explore this issue, and some reporting of this in the Results would be normally expected of a DNA barcode paper
Reply: We agree with the comment and therefore have adjusted Table 1 to include data focused on the BINs utilized in this study, including Maximum genetic distance and also information related to the nearest neighbour BIN. The data of GenBank accession numbers presented in the first submission has been amended and will be included as Supplementary material, while also considering the maximum genetic distance noted per species (using the data from the currently analysed specimens).
Comment: BINs are 1st mentioned in the introduction and later reported in the Results & Discussion, but a simple description of what these are, and why they are useful in analyses, and why they are used in your study, is not presented in the draft. Logically this description should be at least reported in the Methods (and potentially with minor passing comments in the Introduction where 1st mentioned).
Reply: More information on BINs has been included in the introduction and in the methods section.
Comment: NJ genetic distance tree and phylogenetic inference reported in the paper: Authors extensively report NJ distance trees in their results, this is standard and widely used in DNA barcode analyses due to the simplicity of the approach and its relative utility for exemplifying terminal species units based on the genetic distance data. But the authors go one step further and report these NJ relationships as phylogenetic relationships. NJ trees are phenograms (dendrograms) of pairwise genetic distances estimated among terminal samples. Over several decades authorities have argued that NJ distance trees are not appropriate for inference of phylogenetic relationships among terminal samples. The primary reason: NJ combines distances obtained from informative characters (in this case nucleotide sites) into a single ensemble metric - NJ cannot account for individual character (in this case nucleotide) homologies that contribute to the reconstruction and inference of a systemic relationship. There is much discussion of this in the literature, and authors should explore this issue, start with DeSalle R and Goldstein P (2019) Review and Interpretation of Trends in DNA Barcoding. Front. Ecol. Evol. 7:302. doi: 10.3389/fevo.2019.00302. Alternatively, there are a minority of researchers who maintain NJ trees are interpretable as phylogenetic reconstructions. If you can cite or provide a justification for your NJ distance trees to be considered as phylogenetic reconstructions, then report that in your Methods. Otherwise, you will need to change all your reported terminologies involving your NJ trees. For example, "phylogenetic relationship" will need to be replaced by "genetic distance relationship"; mention of "clade" will need to be replaced by "cluster" or "node" depending on context used; "phylogram" replaced by "dendrogram"; "phylogenetic tree" replaced by "genetic distance tree" or “phenogram” etc.
Reply: We do agree that NJ is not the best method to use for phylogenetic analyses, however given that the data set used for each tree was small and composed of closely related species then the data set does not reach a level of saturation in genetic differences as usually noted in distantly related species or in datasets with more terminals than characters. Nonetheless, given the concern highlighted by reviewer 1, we agree on changing this to Bayesian inference using the model of best fit as identified through jModel.
Comment: Terminology: In Results section, Monophyly is reported several times in relation to the NJ trees. Despite the above caveats concerning use of NJ trees for phylogenetic inference, it is still acceptable when reporting NJ results to use terms such as "monophyly" when referring to a cluster of specimens which share the same taxonomic description. But I am curious why other similar terms regarding species systematics relationships have not been used in the draft. In almost every subset tree examined there is at least one species which is paraphyletic with respect to a sister species, these are not formally reported as "paraphyletic", rather we are provided with more verbose description that are difficult to follow. For example, in reporting of Fig 3, Nyctobrya muralis has multiple genetic clusters that are poorly resolved in the NJ tree, the species is genetically paraphyletic with respect to N. segunai.
Reply: We adjusted the terminology as indicated.
Comment: Use of 2% sequence criteria for species delimitation: use of a universal 2% distance threshold as criteria for demarcating species, is cited several times in the paper (eg., lines 191-199) and justification for this is based on reports in several papers cited by authors at line 139. But note this issue of universal distance thresholds applied to DNA barcode studies has been EXTENSIVELY discounted by leading authorities in the field including DNA barcoding founders Hebert at al, and the general consensus is that use of a universal set % threshold for species delimitation is prone to false negative and false positive errors due to differing mtDNA coalescent rates and depths among species – new species do not universally evolve at a set uniform rate of genetic divergence as is naively assumed by use of a sequence distance threshold. Essentially a universal threshold is redundant as an empirically applied analytical statistic in DNA barcode studies (refer summary of issue in Collins RA, Cruickshank RH. The seven deadly sins of DNA barcoding. Mol Ecol Resour. 2013 Nov; 13(6):969-75. doi: 10.1111/1755-0998.12046. Epub 2012 Dec 27. PMID: 23280099). So, what is required here is a more cautious use of language nomination of "potentially" novel taxa, where you have large genetic distances between taxa, and additional supportive evidence of morphological character differences, that is the realm of integrative taxonomy used for species delimitation and discovery and there are plenty of useful paper which describe this approach. Authors should not be so forthright with asserting language such as “justifies it being promoted to species level” used at line 198.
Reply: We removed the 2% criterion from the methodology and focused on BOLD BINs as they use the currently accepted algorithm for clustering OTUs.
Comment: K2P (Kimura 2 Parameter) is used to modify pairwise genetic distances in NJ analyses... why did you do this in a DNA barcode study??? (Refer again to the Collins & Cruishank paper and others). In your NJ reporting of pairwise comparisons, be careful using "divergence" when you are actually referring to "distance", in systematics work the former suggests reproductive isolation the latter is non-inferential in regard to reproductive isolation.
Reply: This analyses has been changed from NJ to BI. The terms ‘divergence’ and ‘distance’ were rechecked and corrected accordingly.
Comment: lines 149-150: mention of a BOLD generated distance tree used for inferring Results, but not observable in the draft. Either include it in a Supplementary section, or refer to it as "(data not shown, eg., at line 285 and elsewhere).
Reply: We are indicating that it is ‘data not shown’ as these are from data generated through BOLD TaxonID Tree and images cannot be transferred.
Comment: lines 169-170: report of nucleotide content. I do not see a need for this, it is not reported elsewhere in the draft and has no explained bearing on the outcomes of your work. Normally we would see such reports in comparisons of different loci - or if there is an extreme nucleotide bias not observed relative to earlier studies or of other taxonomic groups. It has no justified reason to be in your draft.
Reply: The mentioned sentence has been removed.
Comment: Lines 180-182: There is comment stating 46% of Maltese island barcodes are novel in regard to mainland conspecifics (Table 1). But I cannot determine in your Table 1 which GenBank accessioned sequences are novel. So, I cannot determine how you came up with a 46% novelty statistic. This could be resolved by simply adding an asterisk * to those accessions which are novel, simple and accountable.
Reply: The original Table 1 had a column of percentage match per sequence so if the match is not 100% then it would mean that the accession number represents a new haplotype. However, given that Table 1 is being amended, we are indicating the number of haplotype variants noted per species and the number of which are newly recorded haplotype variants. The for the individual sequences we left the original table 1 in the supplementary material and indicated that if the matches are not 100% then it means that the sequence was newly recorded.
Comment: Line 322: replace “morphological identification data” with “taxonomic descriptions”
Reply: The sentence has been amended as indicated by reviewer 1.
Reviewer 2 Report
A very useful paper dealing with DNA barcodes of butterflies and moths of Malta.
The MS is interesting from the following three points of view. First, it represents a lot of new data (new records for Malta and new DNA barcodes). Second, it is interesting from the view point of evolutionary biology, since it provides evidence for molecular distinctness of the Maltese fauna. Third, the paper provides a tool for molecular identification of the butterflies and moths of Malta.
But the work is not without some drawbacks.
The authors are talking about a barcode library. But this is not yet a library, because too little of the fauna has been studied. This is rather the first step towards creating such a library.
The article lacks barcodes for ¾ species and many whole families of Lepidoptera. The list of taxa is dominated by common and mass species. Therefore, I would recommend the authors to shift the focus from creating the barcode library to the first steps of DNA barcoding the Maltese Lepidoptera.
A factual inaccuracy: Polyommatus icarus belongs to the family Lycaenidae, not to Papilionidae. Should be corrected.
Minor corrections:
“around 200 being described as synonyms (syn. nov.) each year”
better to say: “around 200 species names are synonymized each year”
Table 1
“haplotype number per species (H),”
Please replace by
“haplotypic variants within the species (H)”
Author Response
Reviewer 2:
We would also like to thank reviewer 2 for the highlighted issues that need to be tackled to improve this manuscript.
Comment: The authors are talking about a barcode library. But this is not yet a library, because too little of the fauna has been studied. This is rather the first step towards creating such a library. The article lacks barcodes for ¾ species and many whole families of Lepidoptera. The list of taxa is dominated by common and mass species. Therefore, I would recommend the authors to shift the focus from creating the barcode library to the first steps of DNA barcoding the Maltese Lepidoptera.
Reply: We amended the statements surrounding the barcode library, indicating that here we are presenting the ‘first outcomes’ (mentioned in the abstract and ‘first steps towards a comprehensive’ (mentioned in discussion). Indicating that this is not the end, but the first part of an ongoing effort.
Comment: A factual inaccuracy: Polyommatus icarus belongs to the family Lycaenidae, not to Papilionidae. Should be corrected.
Reply: Accepted and amended accordingly throughout text when mentioning number of families and species within.
Comment: “around 200 being described as synonyms (syn. nov.) each year” better to say: “around 200 species names are synonymized each year”
Reply: Accepted
Comment: “haplotype number per species (H),” Please replace by “haplotypic variants within the species (H)”
Reply: Accepted
Reviewer 3 Report
The paper presents the first comprehensive DNA barcode library of the Maltese Lepidoptera. It is well written and presents interesting and useful results. The authors detected several intriguing taxonomic problems which may be solved in further studies (I would recommend to do that).
I have only a couple of suggestions and also found some minor technical errors, which the authors may find useful, and may consider or not them. These are embedded in the file attached.
Comments for author File: 
Comments.pdf
Author Response
Reviewer 3:
We would also like to thank reviewer 3 for the constructive comments in improving this manuscript.
Comment: You may add: "...147 species (ca. 25% of lepidopterous Maltese fauna)..."
Reply: We added this comment in the abstract.
Comment: In this paragraph, I suggest a short note claiming that barcoding is not a universal tool, i.e. discrepancy between barcode, nuclear genes and taxonomic consensus have been reported.
Reply: We added a statement in the second paragraph of the introduction indicating that DNA barcoding is not universal.
Comment: Font Italic
Reply: Corrected
Round 2
Reviewer 1 Report
Authors have provided extensive editing and addressed all issues I raised in my earlier review, to my satisfaction
